Meet Deepansh,
the |

• Led development on an oncology application for deep survival analysis — Plot Digitisations, IPD Reconstruction, Parametric, Cure, Spline, Piecewise, Dependent and Independent Modelling.
• Delivered a vaccine sales tracking app across UK vendors and chains — map-enabled, region-based, millisecond-optimised geospatial loading.

• Led an LLM-powered application for researchers to query the Socialstyrelsen database through natural language. End-to-end LLM chain via ellmer and shinychat with guardrails and interpretability.

• Promoted best practices in R, modular programming, Docker deployments across Render and DigitalOcean.
• Facilitated SaaS products via Stripe and Auth0 integrations into RShiny.

• Developed a generalised SPC package and companion app. Automated manual reports — 1 week down to 2 hours.
• Built a high-frequency monitoring tool to process high-fidelity data using ETL pipelines built in {maestro} in R to mark anomalies and critical failures in electrodes automatically.
• Led multiple Git workshops and standardised CI/CD practices using Azure Pipelines for data scientists, creating several templates in quarto and RShiny to help them write and deploy production-ready code.
• Automated routine tasks in Azure DevOps using the official API and a custom-built R package, saving hours of manual effort to initialise repositories, create pipelines, and more.
• Built LogAnalyzer — an RShiny log-monitoring tool for Posit Connect with optional LLM analysis. Honorable Mention, Posit Shiny Contest 2024.
• Co-created and maintained shared internal packages under the elkemVerse ecosystem — elkemApp, elkemDocs, elkemPlots, elkemSPC, elkemDbConnect, elkemAPI — ensuring 100% unit test coverage for all packages.

• Talks and workshops at ShinyConf25, ShinyConf24, EARL24, useR!24, R/Pharma24.
• Tech talk ambassador — internal sessions and knowledge-sharing.

• Migrated legacy applications to a modular rhino architecture. Refactored large monolithic Shiny apps for maintainability and scalability. Created elkemDashboard, a bslib-styled template that brought consistency to the apps.
• Co-created, contributed to, and maintained shared internal packages under the elkemVerse ecosystem such as elkemApp, elkemDocs, elkemPlots, elkemSPC, elkemDbConnect, elkemAPI, ensuring 100% unit test coverage for all packages.
• Authored a case study outlining the architecture and business impact, acting as a liaison between Appsilon and Elkem.

• Refactored a legacy app, with modules sized at 2500 LOC, to smaller, more manageable 200 LOC modules.
• Enhanced reactable tables with editable rows, YAML-based configs, and tooltips. Implemented report generation and email automation workflows using quarto.

• Collaborated with the Polish Academy of Sciences on an RShiny dashboard using leaflet, echarts4r, and rhino.
• Tech lead — Scrum, planning, retrospectives, delivery.
• Featured at ShinyConf23 Shiny Apps Showcase.

• Moderated AI and LLM panels at ShinyConf23.
• Internal journalist and tech marketing champion.

• Developed RShiny apps for econometrics analyses like Conjoint, MaxDiff, and Forecasting to showcase students how the different parameters interact with each other.
• Facilitated courses on Deep Learning, Big Data and Data Collection, helping professors with grading and assignment planning, assisting students during office hours.
• Conducted tutorials on building end-to-end data streaming pipelines on AWS, showcasing how S3, Kinesis, Kinesis Firehose, Lambda, DMS, and Athena come together to create an end-to-end solution.

• Aided Townscript’s data-driven design, working along with the design and product teams to improve the KYC flow with clickstream tracking, increasing KYC success rate to over 90%.
• Built a custom spam classifier using Latent Dirichlet Allocation (LDA) to combat invalid and abusive events, promoting a feature to mark events as spam while gathering more data for fine-tuning.
• Built a topic architecture from scratch for the B2C events platform employing topic modelling along with in-person and remote user studies to understand user behaviour and expectations.
• Created Townscript’s parquet data lake on AWS S3 by bringing data from within and outside AWS services such as RDS, DMS, Kinesis, MixPanel, Google Analytics, using a customised boto3 script, reducing transfer costs by 96%.
• Worked with the product team to define key metrics, creating asynchronous scripts and dashboards in Python.

• Mentor and reviewer for Machine Learning and ML Engineer Nanodegrees.
• 1-on-1 guidance via chat and calls. 4.73/5 rating.

An MIT-licensed, open-source, self-hosted tool for writers and publishers who want to publish in an aesthetically pleasing website, who ignore the bells and whistles of the modern internet, and who want to create a better experience for their readers. Opinionated, minimal, and easy to use, guided by its own ethos using React and Vite, enabled by Docker and Docker Compose. Facilitates a version-controlled, markdown-based writing and editing paradigm with Git-tracked repositories and an automated two-pane Reader Mode.

Landing · GitHub · Product Hunt

An MIT-licensed, writer-first terminal writing app built with Python and Textual. Minimal, distraction-free writing right in the terminal — no GUI, no clutter, just words. Designed for writers who live in the command line and want their writing tool to stay out of the way. Supports markdown, auto-save, and a clean monospace aesthetic.

Landing · GitHub · Product Hunt

An RShiny log-monitoring tool for Posit Connect with optional LLM-powered analysis. Parses and visualises server logs, helping teams debug deployments and track app health. Built during my time at Appsilon. Honorable Mention, Posit Shiny Contest 2024.

GitHub

An open-source Plumber API for personal metrics such as finances, travel itineraries, calendar, to-dos, and more. Several services such as AWS EC2, S3, DynamoDB, Supabase PostgreSQL, ShinyProxy, DigitalOcean, pm2, Docker were used to build a full end-to-end personal tracking system. Created an accompanying set of dashboards in RShiny, fully mobile-friendly and styled using custom Sass stylesheets.

Presented and released the open-source version at ShinyConf24, useR!24 in Salzburg, Austria, and EARL24 in Brighton, UK.

GitHub

Over 400k words across 1000+ pieces of prose. A collection of unabridged, unedited, and unapologetic public journal entries about life, people, society and everything in between. Also includes vignettes dubbed “bookmarks” and long-form journal entries about one idea or concept. Made to feel like the perfect reading experience without any distractions.

Technical notes and writing on code, tools, and craft.

Nudge › How — a humane approach to self-improvement (2020–2022)
The Polymath — 200+ page personal blog, a journal of explicit growth (2011–2020)
On This One Trip — intimate travel stories, not top-5 lists (2017–2018)
Monologues in a Metro — inner monologues walking through Delhi (2016)
Age of the Pixelator — desktop wallpapers with my cousin (2010)
DBZ Naruto Sprite — sprite animations uploaded from a cyber cafe (2009)
What Next — school journal with code names for classmates (2009)
My first blog — deepanshcool.blogspot.com (2008)

Sticker collection that made it into Counter-Strike: Global Offensive at age 19. Self-taught through years of Photoshop and Illustrator since 7th grade.

Sticker Capsule

Minimal illustration brand inspired by real visits to places. Postcards, canvas prints, and notebooks sold through pop-up shops and an online store. Ran for about a year before closing down at break even. Born on a random evening in 2020.

Illustrations for the self-improvement platform. Minimal — only what is necessary.

Instagram

Started in high school. First dollar and first thousand dollars on Fiverr in 2014, at 18.

Psychology, philosophy, economics. Fiction over non-fiction. Camus, Hemingway, Bukowski, Kafka. Favourite book: The Little Prince.

Minimalism · Reducetarianism and vegetarianism for five years · Absurdism · Finding balance